Make exp() work with non-simple arguments

[arsenovic]

Make exp() work with non-simple arguments

[tBuLi]

Thanks for reporting this. The problem is that this is very hard to do in an input agnostic way, in particular because of symbolic multivectors, and less so for array types like numpy or pytorch. This is also tangentially related to #107, in that it requires a general paradigm for dealing with such functions.

Perhaps this is a good time to have a discussion about this. I think it might be good to have a special module for functions that specialize in particular types, which is fine as long as it is clear that it is a specialization. Thinking out loud, perhaps we could do something like

from kingdon.numerical import exp

z = exp(my_numerical_multivector)

[arsenovic]

works for me

[arsenovic]

this seems like a bug in the broadcasting?

import numpy as np
from numpy import pi 
from kingdon import Algebra
exp = lambda x: x.exp()
 
pga = Algebra(3,0,1,start_index=0) 
locals().update(pga.blades)
 
R  = lambda theta, phi : exp(-theta*e12/2)*exp(-phi*e13/2) 
thetas = np.linspace(-pi,pi,21)              
phis   = np.linspace(-pi/2,pi/2,11)[1:-1]  
theta, phi = np.meshgrid(thetas,phis)
Rs     = R(theta,phi) 

this snippet works without broadcasting (looping over phi,theta), and also works with broadcasting but with the numerical exp().